Skip to content

Releases: CellMigrationLab/CellTracksColab

CellTracksColab 1.1.0: run it in Colab or on your own computer

Choose a tag to compare

@IvanHCenalmor IvanHCenalmor released this 01 Oct 12:24

CellTracksColab is no longer Colab-only. The same notebooks now run in Google Colab and locally, either through a one-click desktop app for Windows, macOS and Linux or in your own conda environment.

🖥️ New: local use

  • Desktop app. Installers for Windows (.exe), macOS (.pkg/.sh, Apple silicon and Intel) and Linux (.sh) are attached below. They bundle Python, JupyterLab and all dependencies, so there's nothing to configure. The app is built with LabConstrictor. See the installation guide.
  • Welcome dashboard. It lists every notebook, opens each with one click, and checks for notebook and code updates.
  • Your own conda environment. Use environment.yaml + requirements.txt + pip install -e .. See Running CellTracksColab locally.
  • Notebooks that work in both places. Colab-only forms were replaced with interactive widgets and Submit buttons. The setup cell detects Colab or local automatically. When running locally, you can choose folders with a native folder selector on Windows and Linux.

📊 Analysis improvements (since v1.0.1)

  • Forward Migration Index (FMI) added to the track metrics.
  • Axis scaling for origin-normalised track plots (x_scale / y_scale).
  • Track plots now have an inverted y-axis that matches the image orientation, and more robust plotting functions.
  • Categorical colour palettes in box plots and dimensionality-reduction plots, so repeats are easier to tell apart.
  • Infinite values are converted to NaN before analysis. Columns with too many NaNs are handled in the plotting and statistics steps.
  • Compressed output. Compiled dataframes are saved as .csv.gz to reduce file size.
  • Filtered dataframes are now saved to the Results folder.
  • Fixes:
    • p-value formatting and heatmaps are more robust (no errors on NaN or infinite values);
    • an empty cluster or condition no longer stops the cluster plots, and missing columns are skipped;
    • the duplicated "Min Speed Rolling Rolling" metric name is fixed;
    • the file-format pattern used when loading files is safer;
    • general bug fixes to the TrackMate notebook.

🗂️ Repository changes

  • Notebooks moved to notebooks/<NotebookName>/. Each one has its own requirements.yaml and CHANGELOG.md.
  • The celltracks package moved to src/celltracks/.
  • The documentation (README, wiki, install guide) was updated to cover both the Colab and local options.
  • Notebook versions are now 1.1.8 (Dimensionality Reduction 1.1.10).

📦 Downloads

Platform File
Windows CellTracksColab-1.1.0-Windows-x86_64.exe
macOS (Apple silicon) CellTracksColab-1.1.0-MacOSX-arm64.pkg / .sh
macOS (Intel) CellTracksColab-1.1.0-MacOSX-x86_64.pkg / .sh
Linux CellTracksColab-1.1.0-Linux-x86_64.sh

Prefer not to install anything? Open any notebook in Google Colab from the README.

v1.0.1

Choose a tag to compare

@guijacquemet guijacquemet released this 29 May 10:56
0b3bb6e
  • Includes a general data reader that supports new data formats
  • New and improved plotting options including collapsing menu, ttest, and log color coded heatmaps
  • Track metrics can now be computed using rolling windows
  • New options to plots the tracks
  • Improved filtering functions
  • various bug fixes
  • Full code review and most of the functions are now externally packaged
  • Plotting functions are imported from the main code

Release v0.9.1

Choose a tag to compare

@guijacquemet guijacquemet released this 06 May 12:07
acfdf49

This is the final release prior to a major update of the platform. This version has been archived for reference purposes.

preprint v3

Choose a tag to compare

@guijacquemet guijacquemet released this 24 Jan 19:44
a917302

Release associated with the latest version of the preprint.

v.02

Choose a tag to compare

@guijacquemet guijacquemet released this 07 Nov 09:51
b0fadef
  • Documentation added
  • many new options and bug fixes

v0.1

Choose a tag to compare

@guijacquemet guijacquemet released this 20 Oct 10:52
284da57

First code archive